VISUALIZING AND CLUSTERING FAKE JOB POSTINGS: DATA-DRIVEN INSIGHTS FOR FRAUD DETECTION

Abstract

Online job platforms have made it easier to find jobs, but they have also made it easier for scammers to post fake job postings, posing risks to job seekers. These fraudulent activities can lead to severe consequences, such as identity theft, financial loss, and emotional distress for victims. To improve recruitment platform security and safeguard users, it is essential to spot trends in these fake job postings. This study focuses on visualizing patterns within fake job postings through data-driven insights, employing various data visualization techniques to reveal key attributes associated with fraudulent activity. A dataset contains both legitimate and fraudulent job postings. Exploratory data analysis (EDA) is conducted to examine variables including salary category, job function, industry, location, and other related features by using categorical distribution, geographical distribution, and word cloud. This study provides insights for recruitment platform controllers, raises user awareness, and facilitates the early detection of fraudulent job posts by displaying clear and actionable visual patterns. The results highlight how visualization and clustering are used to gain insight into characteristics of fraudulent job postings, like the fraudulent job postings predominantly target customer-facing roles in industries like Oil & Energy and Customer Service, which are concentrated in the United States (especially Texas and California), and rely on vague language and unrealistic promises. These findings contribute to more targeted fraud detection strategies and create safer online job search environments.